Nobody on the ETL main stage argued for cheaper
We put one question to the COO of Terrapinn, the CEO of Hemming Group and the head of digital experience at Informa Festivals. Should AI make your events cheaper to run, or worth more? Watch the session, or read what they said.
Full session · 44 min · Event Tech Live, London, 11 November 2026
Sharon RoessenChief Operating Officer, Terrapinn (LinkedIn, opens in a new tab)
Robin BoothChief Executive, Hemming Group (LinkedIn, opens in a new tab)
Cindy AtinaultVP Digital Experience and Operations, Informa Festivals (LinkedIn, opens in a new tab)
Maanas MedirattaHost · Co-founder, Bridged (LinkedIn, opens in a new tab)
Every event business is spending on AI and not everyone can say what it bought them. When we asked the main stage audience at ETL who had spent more than four figures on AI, about half the room raised a hand. Most of them kept it up when asked if they knew the value it created. On stage, the answers were more careful.
We framed the session around one choice. AI can lower your cost to serve, or it can make the event worth more to the people in the room. The answer shapes what you buy, who owns it and how you report it. Here’s what three people running it at scale said, in their own words.
Nobody started with cost
Terrapinn put AI in a director’s job title in 2025 and started with revenue, not savings. Robin Booth, four months into running the 130-year-old Hemming Group, is funding AI partly by not renewing legacy systems. That’s a reallocation, not a cut, and there’s a structural reason the cost case stays weak.
I immediately said there are no cost savings at all. This is a cost increase. Then I realised what I’m probably doing is swapping spend out. I’m funding some of the AI spend by not spending on legacy systems.
If it’s successful, it’s going to cost you more, and it’s very hard to budget until you’ve tried it.
AI shows you the process. It doesn’t fix it
Terrapinn put call intelligence across 150 sales reps. After 16,000 calls, the value wasn’t in the tool. It was in what leadership could now see and change.
You can’t just bolt on AI and expect it to improve. What it’s done is made everything very much more transparent, and blown open the holes in process.
Two questions
First-time visitors to Informa Festivals’ sites couldn’t tell which pass was right for them, and some never decided. Now the site asks job title and company, then returns the right pass and why, a personalised agenda and the people worth meeting. Fewer touchpoints to a decision, and a visitor who arrives with a plan.
Personalisation has been on the marketing wish list for years. On the website it’s very hard, because you need to bring the data and you can’t scale it. AI in this area was a game changer for us.
The £17,000 question
A Bridged workflow on one of Hemming’s exhibition sites found a lead that became a £17,000 stand booking, on a new floor the team hadn’t sold before. Robin’s own first reaction was to question it. Was that business coming anyway? At minimum it was found faster, but even a black and white result can’t be fully separated from everything else that changed.
Even when you have the clearest sense of what looks like a black and white ROI proof, you’re still not a hundred percent sure.
Cindy’s team gets closer with two methods: a blocker survey on the pass page, which should fall as the tool does its job, and a controlled test of visitors with the tool against visitors without it.
Broadcast is getting harder, partly because of AI
Inboxes are now filtered, sorted and sometimes read by AI before a person sees them. Reply rates and deliverability are falling. The panel kept coming back to conversation over broadcast: fewer, more relevant messages on channels people answer. Robin’s tip was to watch WhatsApp, because consumer behaviour changes first and Amazon now takes orders there.
The request I get is the name of the company and the name of the person. It’s assumed it’s then on me to work out who that company is and whether their product is relevant to me. We can do better than that as an industry.
Users, not builders
Every failure the panel shared came back to data or people, not technology.
People generally are users, not builders. You need to build the use case. Connect the data first, build them a use case, then teach them, and keep going.
Putting it live is the easy bit. Getting customers to understand what you’ve done, why you’ve done it, and how it benefits them is the hard bit.
If you’re not failing, it means you’re not outside your comfort zone.
What we’d measure
As AI floods the internet, a room full of people becomes more valuable, and harder to fill. Our view after the session: at the level of a single workflow, cheaper or worth more is a false choice. Pick one problem, agree the KPI before anything goes live, and measure it against a baseline.
One piece of advice each
Stop trying to fix everything.
The only failure would be doing nothing.
Start with one use case. Measure it. Then scale.

